The present invention relates to a distortion compensation circuit for improving secondary distortion characteristics in an optical transmitter and a high-frequency amplifier used in a transmission path for a broadband high-frequency signal such as a cable television (CATV) etc.
In the initial CATV system, all of the transmission lines from a head end to respective subscriber's terminals are formed by coaxial lines. Further, in order to compensate signal loss of the lines, amplifiers are provided usually with an interval of about 500 m, that is, an interval of about 20 dB of line loss, whereby the signal is transmitted such that it is repeatedly and alternately attenuated and amplified.
The number of stages of the amplifiers connected in series in this CATV transmission line differs depending on the scale of the facility and the loss performance of the cable to be used, and becomes thirty (30) or more when used in a large scale facility. In this case, the signal is added with noise and distortion component each time passing through an amplifier and hence the quality of the signal degraded gradually.
In such a CATV system, the signal reached a subscriber's terminal after passing through maximum number of amplifiers in the CATV facility is required to satisfy the signal quality condition defined by the technical standard of the rules of the cable television broadcast law, for example. Further, the efficiency of the amplifier etc. should be determined and distributed so as to at least satisfy such a condition.
Although the signal quality of the CATV is legally regulated in this manner, it can not be concluded that the efficiency of the facility is sufficient so long as such a condition is satisfied. This is because there is a background that the technical standard was determined based on the signal quality required in facilities at which it is intended to eliminate such a problem by using cables that the facilities exist in areas with poor signal reception such as places among mountains where high-quality signal reception can not be. Further, this is because the signal fault caused by noise and distortion component is liable to be detected more easily due to the fact that recently the screen size of television receivers has been increased and the quality of image of video appliances has been improved.
In view of the aforesaid circumstances, the signal quality required in the actual CATV facilities is higher than the legally regulated value. The degree of such signal quality has been presented in the form of the conferences and recommendations of the industrial associations etc. and in the form of the individual required standards of the respective CATV corporations, and the contents of the required quality has been increased year after year. This fact means that, in the case of a facility using only coaxial lines, the number of stages of amplifiers can not be increased to a large number and hence the scale of the facility is limited.
As a means for eliminating the limitation of the scale of such a CATV facility using only coaxial lines, the hybrid fiber coaxial (HFC) system has been recently put to practical use.
In the HFC system, all-channel high-frequency signals to be transmitted are transmitted from an optical transmitter (E/O) in a head end to an optical receiver (O/E) through an optical fiber, then the signals are converted into the high-frequency signals by the optical receiver and distributed to subscribers through coaxial lines.
This HFC system is able to transmit signals over the distance of about 20 km at maximum between the E/O and the O/E without using any repeater but through an optical fiber line which is substantially lower in loss as compared with a coaxial line. Thus, it is possible to provide a CATV facility having a scale sufficient for covering the entire area of a single city, for example, by laying a plurality of optical lines therein. Such a distance corresponds to 40 stages of serially connected amplifiers of the coaxial system. If such a CATV facility is formed only by coaxial lines, the signals will be degraded to no practical degree. In the HFC system, the number of stages of amplifiers constituting the coaxial system at the succeeding stage of the O/E is usually limited to about 6 so that the degradation of signal quality is suppressed at this portion.
In order to operate the HFC system effectively, the efficiency of the optical transmission system thereof is required to be better than that of the coaxial transmission system formed by several ten stages of amplifiers for transmitting a signal over a distance equal to that of the optical transmission system.
The factors for determining the quality of the CATV transmission system are a carrier to noise ratio (C/N) and distortion characteristics. The distortion characteristics is classified into secondary distortion component and third order distortion component respectively defined as composite second order (CSO) distortion and composite third order distortion (composite triple beat: CTB). The reason for calling such distortions in this manner is as follows. That is, since the carriers of the television signals are disposed with an equal interval of 6 MHz on the frequency axis, the distortion components dropped into the frequencies obtained through the addition and subtraction of the frequencies of the respective carriers appear intensively at the frequencies which exist with almost constant interval from the respective carriers despite that the distortion components accompany the frequency deviations of the respective channels (allowable deviation under regulations is within 20 kHz). Thus, the distortion components appear as composite of many distortions slightly differing in frequencies. Such distortions are detected as beat of striped pattern on a screen.
Other fault caused by the third order distortion is a fault called as cross modulation in which the modulation contents of other channel overlaps on a screen. In a CATV facility having transmission channels about more than 20 channels, that is, in most CATV facilities, the aforesaid composite beat fault appears mainly and the cross modulation is negligible. This is because, although the degree of the cross modulation increases simply in proportional to the channel number, the degree of the CTB increases almost in proportional to the number of the combinations employing 3 from the channel number n, that is, nC3.
Among the aforesaid factors for determining the quality of the CATV transmission system, the optical transmission path can easily exceed in the C/N and CTB characteristics than the coaxial transmission path when the lengths of both the paths are same, in particular, due to the recent improvement of the efficiency of the laser diode (LD) used as an optical source of the optical transmitter.
However, the efficiency of the optical transmission path is almost same in the CSO of the secondary distortion as that of the coaxial line having the same length as that of the optical transmission path. As a matter of course, the C/N characteristics and the distortion characteristics are in an opposite relation, that is, when the modulation degree is made larger, the C/N characteristics can be improved but the distortion characteristics is degraded. When the C/N characteristics and the CTB characteristics are determined at the compromise point therebetween, the CSO becomes insufficient level. In contrast, when the CSO and the CTB characteristics are determined at the compromise point therebetween, the C/N characteristics becomes insufficient level.
This fact derives from the nature of the LD. In the case of the high-frequency amplifier constituting the coaxial system, the secondary distortion component can be easily suppressed by forming the circuit arrangement in a push-pull type. In contrast, in the case of the LD, such an operation can not be expected and the secondary distortion with a higher level than the third order distortion will be generated.
Although the optical transmission path is superior than the arrangement of the all coaxial line when the lengths of both the paths are the same, the HFC type is not sufficient in the signal quality due to the bad secondary distortion characteristics peculiar to the LD. The HFC type would be able to provide high grade service to subscribers if such a problem is obviated.
The predistortion method has been proposed as a means for improving such distortion characteristics. This method makes the entire operation linear thereby to cancel the distortion component by adding, at the pre-stage of an element having non-linear characteristics such as an LD etc., a circuit having non-linear characteristics reverse to that of the element.
FIG. 2 is a block diagram showing the configuration of a conventional distortion compensation circuit. U.S. Pat. No. 5,132,639 (issued on Jul. 21, 1992) is raised as the conventional example and the operation thereof will be explained with reference to FIG. 2 (FIG. 1 in the U.S. patent).
The distortion of reverse characteristics is added between a high-frequency signal input 12 and a terminal 25 to a transmission device. A light emitting diode (LED) or a semiconductor laser, that is, a laser diode (LD) is mentioned as the transmission device, but none is shown in figures.
The electric power of the high-frequency signal applied to the high-frequency signal input 12 is partially branched into a path 14 by a directional coupler 10 and sent to a distortion generation unit 15. The distortion generation unit 15 generates the secondary distortion or more higher order distortion (this U.S. patent intends also to cancel the third order distortion). The distortion component thus generated is passed through an amplitude adjusting unit 17, a frequency tilt adjusting unit 19 and a phase adjusting unit 21, then applied as a distorted signal 22 to a directional coupler 11, in which the distorted signal is mixed with a delayed modulation signal 24 which is subjected to only time delay without being distorted. In this case, a ratio of the amplitude of the distortion component relative to that of the modulation signal and the phase relation therebetween are adjusted so as to be reverse to those of the distortion characteristics of a laser diode connected to the rear stage of the directional coupler 11.
The high-frequency signal partially branched into the other path 13 by the directional coupler 10 is delayed by a delay time adjusting unit 23 by a time corresponding to the delay time of the distortion generation unit 15 side having a longer transmission path than the delay time adjusting unit 23 side. Then, the output signal of the delay time adjusting unit 23 is supplied to the directional coupler 11 as the delayed modulation signal.
Explanation has been made as to the configuration and the operation theory of the conventional distortion compensation circuit. The conventional distortion compensation circuit has a problem that the uniform distortion canceling condition can not be attained as to all the signals distributed in a wide frequency band. This is because the conventional distortion compensation circuit employs many high-frequency transformers as the constituent elements thereof. Further, such a problem has become more difficult to obviate since the frequency band of the CATV has expanded from the conventional lower limit of 70 MHz and upper limit of about 300 MHz to the upper limit of 450 MHz, and recently to the upper limit of 750 or 800 MHz and so the number of channels have been increased.
FIG. 3 is a vector diagram for explaining the required characteristics of a distortion compensation circuit. Supposing that the reverse distortion component having the same amplitude as the distortion generated by a laser diode can be generated and supplied to the laser diode, allowable amounts of phase error or delay time error of the reverse distortion component will be obtained as follows in accordance with FIG. 3.
A vector OA is one representing the distortion component generated by a laser diode. A vector for completely canceling the distortion component is one denoted by OAbar (Abar represents a reversed sign of A and a horizontal line is added thereto in FIG. 3) having the same amplitude as the vector OA and having a phase deviated by 180 degrees therefrom.
An allowable angle deviation .theta. of a vector OB with respect to the vector OA will be calculated under a condition that the required improvement factor of the distortion is 10 dB and the vector OB is same in amplitude as the vector OA and the phase difference thereof from the vector OA is not accurately 180 degrees. The improvement factor of 10 dB means that the magnitude of the cancellation remaining vector resulted from the addition of the vectors OA and OB is 1/.sqroot.10 times, that is, about 0.316 times as large as the magnitude of the vector OA or OB.
That is, supposing that the magnitude of the distortion component OA is 1.0, the magnitude of the reverse distortion component OB is 1.0 and the magnitude of the cancellation remaining vector OP is 1/.sqroot.10=0.316.
The magnitude of the angle .theta. is obtained as follows under such a condition. EQU .theta.=180 degrees-2.times.arccos {1/(2.times..sqroot.10)}.apprxeq.18.2 degrees.
In other words, the vector of the reverse distortion component vector is required to be held within 180 degrees.+-.18.2 degrees with respect to the distortion component. This allowable limit angle is represented as follows as an allowable time difference .DELTA.t at 800 MHz which is considered as the upper limit frequency of the current CATV system. EQU .DELTA.t={1/(800.times.10.sup.6)}.times.(18.2 degrees/360 degrees)[seconds]=63.2[picoseconds]
That is, in order to maintain the distortion improvement factor of 10 dB at the frequency band of 800 MHz, the delay time deviation of the circuit is required be within .+-.63.2 picoseconds. The allowable time difference thus obtained is a value at the upper limit frequency of the frequency band. The allowable time difference is considered to be larger in inverse proportional to the frequency in the lower frequency range when the allowable limit angle is same.
However, in the case where, for example, two signals of about 400 MHz are added in their frequencies due to the secondary distortion and drops as the distortion around 800 MHz, the phase differences of these two signals will be added. That is , the angle difference almost half of the aforesaid allowable angle deviation of .+-.18.2 degrees is allowed.
In other words, the delay time difference determined by the upper limit of the frequency band must be maintained over the entire range of the frequency band. In view of the aforesaid nature of the allowable phase difference or the allowable delay time difference, it will be understood that the presence of a high-frequency transformer results in a main cause of the error. In the arrangement of the conventional distortion compensation circuit shown in FIG. 2, each of the directional couplers 10, 11, distortion generation unit 15 etc. includes as a constituent element a high-frequency transformer which is formed by winding wires around a core formed by ferrite.
FIG. 4 is a measured example showing the delay time characteristics between the input and the output of the directional coupler 10 disposed between the high-frequency signal input 12 and the output 13 of the conventional distortion compensation circuit in FIG. 2. In FIG. 4, the frequency is plotted on the ordinate from 0 MHz at the left end to 1000 MHz at the right end in a manner that one scale represents 100 MHz, while the delay time is plotted on the abscissa in a manner that one scale represents 100 picoseconds. The ".gradient." shaped sign on the plotted curve is a maker representing 800 MHz which is set to be the center or reference point on the abscissa.
This relation of both the ordinate and the abscissa is common to the respective figures (FIGS. 5 and 6) showing the following measured data. The example of FIG. 4 exhibits the substantially flat delay time characteristics between 70 to 800 MHz in the CATV frequency band and the delay time deviation is kept within the aforesaid required value of .+-.63.2 picoseconds.
FIG. 5 is a measured example showing the delay time characteristics between the input terminal and the branched terminal of the directional coupler 10 disposed between the high-frequency signal input 12 and the branched output 14 of the conventional distortion compensation circuit in FIG. 2.
In FIG. 5, the delay time varies by about 100 picoseconds at maximum between 70 MHz and 800 MHz. Although each of the delay time characteristics shown in FIGS. 4 and 5 relates to the single directional coupler 10, the arrangement of the conventional distortion compensation circuit includes the directional coupler 11 similar to the directional coupler 10 at the signal output side.
Supposing that another directional coupler having the same characteristics as the directional coupler 10 is employed as the directional coupler 11 such that it is disposed in reverse direction with respect to the directional coupler 10, there will be the delay time variation with respect to the frequency like FIG. 5 between the branched terminal 22 and the output 25 and hence the total delay time variation will be about 200 picoseconds. This total delay time variation largely exceeds the aforesaid required delay time variation range of .+-.63.2 picoseconds. 200 picoseconds corresponds to the phase difference of about 60 degrees at the frequency of 800 MHz. If this phase difference is applied to the angle .theta. of the vector diagram shown in FIG. 3, the respective vectors OA, OP, OB form the respective sides of an equilateral triangle, so that the lengths of these vectors become same, which means that the no cancellation of the distortion will be made.
FIG. 6 is a measured example showing the delay time characteristics of a 180 degree splitter 30 shown in FIG. 7 (corresponding to FIG. 2 of U.S. Pat. No. 5,132,639) which is an example of the distortion generation unit 15 of the conventional compensation circuit of FIG. 2.
This is a high-frequency transformer for performing, in brief, unbalanced state to balanced state conversion, which generates the balanced state in which both electrodes have positive and negative voltages symmetrical with respect to the ground from the unbalanced input in which one electrode is always grounded and so at a zero voltage. Since the measuring system is the unbalanced state, this data is obtained by connecting two transformers of the same condition so as to oppose their balanced side to each other thereby to perform the unbalanced state--balanced state--unbalanced state conversion. Accordingly, half of the measured value becomes the delay time characteristics of a single transformer. The measured variation of the delay time characteristics is about 160 picoseconds for two transformers and about 80 picoseconds for one transformer in the frequency range of 70 MHz to 800 MHz, and further surges in a wave shape periodically with respect to the frequency.
As clear from the aforesaid description, according to the configuration of the conventional distortion compensation circuit, the circuit system for adding the distortion contains at least the two directional couplers 10, 11 and the high-frequency transformer in the distortion generation unit 15. The accumulated amount of the variation of the delay time of such constituent elements largely exceeds the allowable variation of the delay time discussed above.
Of course, the conventional arrangement shown in FIG. 2 includes portions for adjusting the amplitude, frequency tilt and phase, and so it is considered that these factors can be corrected when they change monotonously with respect to the frequency. However, it is impossible to correct such factors within the actual circuit scale in the case where the changes occur in a narrow frequency range periodically like the delay time characteristics. Further, since it is clear that the amplitude, frequency tilt and phase can not be controlled independently without influencing on remaining ones of these factors, it is actually impossible to adjust the amplitude as well as the phase so as to be reverse to the distortion characteristics of the LD appearing in a wide frequency band.
Accordingly, although the conventional distortion compensation circuit arrangement can be operated so as to improve the distortion characteristics at a particular frequency portion, the conventional distortion compensation circuit has a drawback that the improvement of the distortion characteristics can be scarcely obtained or the distortion characteristics is degraded at other frequency portion.
Further, the high-frequency transformer is usually formed by winding for several turns a wire with a diameter of about 0.1 to 0.3 mm, whose surface is subjected to insulation processing, around a core material having holes with a diameter of about 1 mm. Thus, the high-frequency transformer has a problem that, since the delay time characteristics thereof changes slightly depending on the piled order of the wire which can not be controlled and the degree of slack of the piled wire, the adjustment of the delay time characteristics can not be expected to be realized with good reproducibility.
Furthermore, the conventional distortion compensation circuit includes many variable resistors and variable capacitors intended to separately control amplitude, phase or the like in addition to many high-frequency transformers and further includes amplifiers (32, 33 in FIG. 7) as parts for generating distortion. Therefore, the conventional distortion compensation circuit has a problem that the circuit arrangement is complicated, large electric power is consumed, adjustment takes a long time and so the cost is expensive.